Great conference throughout the years.
I think there was a failure of imagination, though.
When numbers dwindled, why not do what other top tier systems conferences do - move outside the US and Bay Area!?!?
I know the answer as to why it didn't. A place on the PC or steering committee became so important for tenure/promotion in the US, that it was 90%+ US people - a social group factor. That led to no attempt to widen the circle of general interest. If they had come to Europe 20 years ago, it would have been huge and breathed new life into Usenix ATC.
PC membership is not necessary for tenure/promotion. Eg my advisor has only been on ~5 PCs in a ten-year career, and has gained tenure at two top-10 CS schools. In general it’s more of a checkmark.
Not sure what you mean by "US people" - the PC is mostly (but definitely not 90%) US based, but that follows from systems research being largely US based (for now, at least).
90% of systems researchers are not US-based. My point is that it didn't embrace global leadership in systems research as a conference. It died because it didn't grow. Eurosys was created as a counter-weight to Europeans not being able to get on the PC of conferences like Usenix ATC. It's still going strong.